Arterial Blood Gas Values Cannot Rule Out PE

MONTREAL -- Physicians should not exclude a diagnosis of pulmonary embolism based on arterial blood gas analysis, according to a study presented at the annual meeting of the American College of Chest Physicians.

"Instead, they should continue to rely on D-dimer and clinical prediction models, such as the Wells criteria, or go on to diagnostic imaging," said Dr. Tara Keays of the University of Ottawa Hospital.

Researchers have tried to come up with bedside investigation tools to rule out this common and lethal disease without the need for more invasive and expensive diagnostic imaging prior to the initiation of anticoagulation therapy.

The most successful results involve blood testing for D-dimer, a fibrin degradation product that is produced only after a clot has formed and is in the process of being broken down, in combination with a clinical algorithm.

Recently, initial studies combining D-dimer with arterial blood gas (ABG) values achieved as a diagnostic method showed impressive negative predictive values, but efforts to validate those results have been unsuccessful.

In the retrospective study presented by Dr. Keays, her team investigated the role of ABG and D-dimer values using data from a double-blind, randomized controlled trial comparing bedside diagnostic tests with ventilation/perfusion (V/Q) scanning in the exclusion of suspected pulmonary embolism (PE).

"In the derivation study, PE was excluded if two of the three bedside studies were negative; the V/Q scan was normal or near normal; or the patient was not started on anticoagulation at the end of the investigation. PE was confirmed by a high-probability V/Q; by low or intermediate probability V/Q along with another diagnostic modality being positive; on autopsy; or if anticoagulation was started by the end of the investigation," Dr. Keays said.

Patients were excluded from the study if they were aged younger than 18 years, had a suspected survival of less than 3 months, were ventilated, had known chronic PE, were on anticoagulation, or had venal caval interruption. Of 824 patients screened, 458 were eligible and 399 consented, and 278 had ABG drawn initially. Of the total cohort of 399 patients, 14.3% were diagnosed with PE.

"Looking at the continuous ABG values, there was no significant difference in the mean PaCO2, PaO2, or Aa gradient between the two groups. However, when we looked at the proportion of abnormal ABGs and D-dimer values, there was a significant difference," Dr. Keays explained.

"Our clinical prediction rule simply states that PE is possible if D-dimer is positive or if the D-dimer is negative and there's an abnormal PaCO2 and an abnormal Aa gradient," she said, noting that the negative predictive value and sensitivity of the rule were 100%, with a true negative proportion of 37.6%, meaning that, in this population, "37.6% of patients could be correctly excluded as not having PE," Dr. Keays said.

A subsequent attempt to validate these findings in a retrospective secondary analysis of 246 patients was not as clear-cut. This study showed a negative predictive value of 91.3%, a sensitivity of 89.2% and a true negative proportion of 30%.

"In the derivation study, it did appear that normal PaCO2 and normal Aa gradient, combined with a negative D-dimer, could exclude PE without the need for diagnostic imaging.

However, in the validation group this clinical prediction rule did not appear to validate. Therefore, normal ABG data in combination with negative D-dimer does not allow safe exclusion of PE without going on to diagnostic imaging," Dr. Keays concluded

"This is an interesting study by Dr. Keays and colleagues to explore whether basic bedside investigation tools can be diagnostic for PE without direct imaging studies such as duplex ultrasound or spiral CT. If their proposed clinical prediction rule was reliable, then the need for more expensive studies could be used more selectively," said Dr. Vivian Gahtan, professor of surgery, University Hospital, State University of New York Upstate Medical University, Syracuse, N.Y. when asked to comment upon this study.

"Their findings show that direct imaging is still an important diagnostic tool, as ABG data and D-dimer results combined were ultimately not conclusive for the exclusion of PE," Dr. Gahtan stated.